Sony FX5 Codec Test: X-OCN LT vs X-OCN C2 vs H.264 (and a ProRes Surprise)

When a new cinema camera lands, everyone talks about the sensor. I wanted to talk about the files. What do you actually get when you switch the Sony FX5 from H.264 to X-OCN RAW? How much detail do you gain, how far can you push the grade, and above all, how much storage does it eat? I shot the same portrait in every codec, graded it once, and measured every file.

Same light, same lens, same grade: only the codec changes.

Where and how I shot it

The footage was shot on 28 July 2026 at the Sony FX5 Touch & Try at TAV Academy in Cologno Monzese (Milan). It was one of the first Italian events where you could get your hands on the camera. Thanks to the TAV Academy team and the Sony specialists for the setup and the time with the camera.

The setup

  • Camera: Sony FX5 (ILME-FX5)
  • Lens: ZEISS Nano Prime 100mm T1.5
  • Settings: 25p, 180° shutter, ISO 800, 4400 K, S-Log3 / S-Gamut3.Cine
  • Codecs tested: X-OCN LT (5K Open Gate 3:2 and 5K 16:9), X-OCN C2 (5K Open Gate 3:2), H.264 10-bit 4:2:2 in camera (UHD 4K and Full HD)
  • Post: DaVinci Resolve Studio, one node tree copied to every clip

Everything was shot as a portrait under the same practical lights: warm key, colourful bokeh and deep shadows. That is exactly the kind of scene where compression shows up first.

The codecs, explained simply

X-OCN LT: the “light” RAW

X-OCN is Sony’s compressed RAW format. You get 16-bit linear sensor data, so white balance, ISO-style exposure and the RAW decode can all be changed in post. LT is the lightest of the X-OCN flavours. In 5K Open Gate 3:2 (4992 × 3328) the FX5 wrote about 763 Mbps. In 5K 16:9 (4992 × 2808) that dropped to about 646 Mbps, simply because there are fewer pixels.

X-OCN C2: even lighter RAW

C2 is still 16-bit X-OCN at the same 5K Open Gate resolution, but at about 459 Mbps. That is roughly 40% smaller than LT. In my test I could not see a meaningful difference in the grade or in the skin detail at normal viewing size. At 400% you can start hunting for differences, but you really have to look for them.

H.264 10-bit 4:2:2: the everyday codec

In camera, H.264 UHD 4K 10-bit 4:2:2 Long GOP came in at about 160 Mbps and Full HD at about 73 Mbps. That is a fraction of the RAW. It still looks very good with a sensible grade, and it plays back on any laptop. The trade-off is that white balance and exposure are baked in, and the Long GOP compression softens fine texture. You can see that on skin and eyelashes in the 400% zoom.

What about H.265?

I didn’t shoot H.265 in this test, but it is what most people will record day to day, so it deserves a mention. H.265 (HEVC) is a more efficient codec than H.264. At the same quality it needs a noticeably lower bitrate, so you get smaller files or more quality for the same size. The price is decoding: it is heavier on older computers, and editing can feel sluggish without hardware acceleration or proxies. If your machine handles it, H.265 10-bit is the best “small file” option for long events. H.264 is the safest choice when compatibility and smooth editing matter most.

Log vs graded: how far can you push it?

For every codec, the video shows a split screen with the ungraded S-Log3 image on one side and the finished grade on the other. Both X-OCN flavours take the grade effortlessly. Skin stays clean, the highlights on the hair roll off nicely, and the shadows hold together. The 10-bit H.264 is far better than people expect, but it is the first to show its limits once you push saturation and contrast hard.

Detail and noise at 200% and 400%

I zoomed in on the eyes, which is where the 100mm was focused, because that is where the codec has the most real detail to keep. X-OCN LT and C2 keep eyelashes, iris texture and fine skin detail crisp. The in-camera H.264 is clearly softer at 400%. That comes from the Long GOP compression and from recording 4K instead of 5K. In the out-of-focus areas the RAW also keeps smoother gradients in the bokeh and less mush in the shadows.

The ProRes plot twist

Many people shoot RAW and then transcode to ProRes for editing. So I exported the ungraded S-Log3 from X-OCN LT, X-OCN C2 and H.264 to Apple ProRes 422 HQ, re-imported the files and applied the exact same node tree.

Lesson 1: keep your RAW decode settings

My first ProRes renders did not match. Part of my grade lived in the Camera Raw tab of DaVinci Resolve (the RAW decode settings), not in the nodes. A plain “ungraded” export throws those settings away. Once I rendered the ProRes with the clip’s RAW settings applied and all the nodes switched off, the same grade matched the original almost perfectly. If you transcode RAW, bake your decode settings in first.

Lesson 2: ProRes is bigger than the RAW

ProRes 422 HQ at 5K came in at about 1474 Mbps, nearly twice the size of the X-OCN LT it came from. The H.264 4K clip grew from 160 to about 611 Mbps once transcoded. Transcoding does not add quality. It only makes files easier to edit, and in this case much bigger.

How much space does it take? Real numbers

These are the average bitrates I measured (file size ÷ duration) from the actual files, plus what that means per hour and per card.

CodecResolutionBitrateGB per minuteGB per hour
X-OCN LT 3:24992 × 3328≈ 763 Mbps≈ 5.7≈ 343
X-OCN LT 16:94992 × 2808≈ 646 Mbps≈ 4.9≈ 291
X-OCN C2 3:24992 × 3328≈ 459 Mbps≈ 3.4≈ 207
H.264 4K 10-bit 4:2:23840 × 2160≈ 160 Mbps≈ 1.2≈ 72
H.264 HD 10-bit 4:2:21920 × 1080≈ 73 Mbps≈ 0.55≈ 33
ProRes 422 HQ 5K (transcoded)4992 × 3328≈ 1474 Mbps≈ 11.1≈ 663

How many minutes fit on a card?

Approximate recording time, based on the nominal card capacity. Real usable space is a little lower.

Codec256 GB512 GB1 TB
X-OCN LT 3:2≈ 44 min≈ 1 h 29 min≈ 2 h 55 min
X-OCN LT 16:9≈ 52 min≈ 1 h 45 min≈ 3 h 26 min
X-OCN C2 3:2≈ 1 h 14 min≈ 2 h 28 min≈ 4 h 50 min
H.264 4K 10-bit≈ 3 h 33 min≈ 7 h 06 min≈ 13 h 53 min
H.264 HD 10-bit≈ 7 h 45 min≈ 15 h 30 min≈ 30 h

To put it in wedding terms: a full day with around 4 hours of footage is about 1.4 TB in X-OCN LT 3:2, about 830 GB in X-OCN C2, and under 300 GB in H.264 4K. With RAW you will need more cards, bigger and faster SSDs for offload and backup, and a proper archive plan. Count every terabyte at least twice, or three times with an off-site copy.

So which codec should you shoot?

Choose X-OCN LT if…

  • you want maximum flexibility in the grade and in the RAW decode (white balance, exposure);
  • you are shooting short-form, commercials, music videos or hero moments;
  • you have the storage and the workstation to handle it.

Choose X-OCN C2 if…

  • you want the RAW workflow at around 40% less storage than LT;
  • you shoot longer formats but still want 16-bit RAW. For most real-world work this is the sweet spot.

Choose H.264 / H.265 10-bit if…

  • you record long events: ceremonies, speeches, documentaries;
  • you need fast turnaround and smooth editing on any machine;
  • you nail white balance and exposure in camera.

My take: C2 for the important moments, 10-bit H.264/H.265 for the long takes. Skip ProRes unless your edit really needs it, because it ends up heavier than the RAW itself.

Watch the full comparison

The full video includes the log vs graded splits, the 200% and 400% zooms and the ProRes test side by side. It is on my YouTube channel, with a 9:16 version on Instagram.

Shot on Sony FX5 with a ZEISS Nano Prime 100mm T1.5 at TAV Academy, Cologno Monzese. Edit and grade by Alex Sofo.

Add a comment...

Your email is never published or shared. Required fields are marked *

Alex Sofo Wedding Films